Prolonged exposure to polluted air and risk of overweight and central obesity in Chinese children and adolescents

Abstract

Background

The rapid economic development in China has led to significant challenges in air pollution. This study examines the relationship between prolonged exposure to air pollution and the prevalence of overweight and central obesity among Chinese children and adolescents.

Methods

A total of 161,093 children and adolescents (aged 9–18) from 32 provinces in China were analyzed. Ambient air pollution (PM_2.5, PM₁₀, SO₂, NO₂, O₃, CO) was assessed using satellite models. Overweight and central obesity were measured via BMI, waist circumference, and Waist-to-Height Ratio (WHtR). Mixed-effects regression models were adjusted for demographic and behavioural factors, with sensitivity analyses conducted for exposure windows.

Results

PM_2.5 was significantly associated with an increased risk of being overweight (OR = 1.212, 95% CI: 1.169–1.257, p < 0.001). SO₂ (OR = 1.103, 95% CI: 1.069–1.139, p < 0.001), NO₂ (OR = 1.069, 95% CI: 1.051–1.087, p < 0.001), and reciprocal CO (OR = 1.052, 95% CI: 1.043–1.061, p < 0.001) also showed positive associations. In contrast, PM₁₀ (OR = 0.954, 95% CI: 0.938–0.969, p < 0.001), AQI (OR = 0.959, 95% CI: 0.935–0.983, p < 0.001), and O₃ (OR = 0.949, 95% CI: 0.935–0.962, p < 0.001) were negatively associated with overweight. For central obesity, AQI (OR = 1.118, 95% CI: 1.095–1.141, p < 0.001) and NO₂ (OR = 1.065, 95% CI: 1.051–1.079, p < 0.001) showed positive associations, while PM₁₀ (OR = 0.901, 95% CI: 0.889–0.913, p < 0.001) and O₃ (OR = 0.973, 95% CI: 0.962–0.985, p < 0.001) were inversely associated. SO₂ and CO⁻¹ were not statistically significant for central obesity after full adjustment. Boys, high school students, and children from low-income or southern regions exhibited stronger associations. Similar trends were observed for central obesity.

Conclusion

Prolonged air pollution exposure significantly increases the risks of overweight and central obesity in Chinese children and adolescents.

Graphical Abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-025-23743-8.

Introduction

Childhood obesity has become a global public health crisis, carrying profound and enduring risks to both physical and mental health [1]. Children and adolescents with obesity are significantly more likely to develop chronic conditions, including cardiovascular diseases, type 2 diabetes, and hypertension, with these complications frequently persisting into adulthood [2]. Moreover, childhood obesity is strongly linked to psychological challenges such as diminished self-esteem, heightened anxiety, and social isolation [3]. The pervasive nature of childhood obesity and its long-term implications for life expectancy and quality of life underscores the urgency of effective prevention and early intervention strategies.

The etiology of obesity is multifactorial, involving interactions between genetic predisposition, behavioral patterns, and environmental exposures [4]. Among these, environmental factors have garnered increasing attention, particularly in light of global urbanization and industrialization. Notably, air pollution has emerged as a potential contributor to the obesity epidemic [5]. Traditionally associated with respiratory and cardiovascular diseases, mounting evidence now implicates chronic exposure to air pollutants—such as particulate matter (PM_2.5, PM₁₀), nitrogen dioxide (NO₂), and sulfur dioxide (SO₂)—in adverse metabolic health outcomes [6–9].

Children and adolescents are particularly vulnerable to the deleterious effects of air pollution due to their developing physiological systems and heightened exposure through physical activity [10, 11]. Emerging evidence suggests that prolonged exposure to elevated air pollution levels may contribute to obesity by inducing systemic inflammation, oxidative stress, and hormonal dysregulation [12–14]. These mechanisms can disrupt metabolic homeostasis, promote obesity, and reduce physical activity levels in response to poor air quality, further compounding obesity risk [14, 15]. For instance, research conducted in highly polluted regions of China has identified positive associations between long-term exposure to particulate matter and NO₂ with increased body mass index (BMI) and central obesity, particularly among boys and socioeconomically disadvantaged populations [16].

To address these pressing concerns, the present study aims to fill critical gaps in the existing literature, including limitations in sample size, methodological rigour, and the precision of air pollution exposure assessments. Leveraging a nationally representative sample of Chinese children and adolescents, this study aims to elucidate the relationship between prolonged exposure to air pollution and the prevalence of overweight and central obesity. The findings aim to establish a robust scientific foundation for targeted obesity prevention initiatives in polluted areas and inform policy interventions addressing air quality and its broader implications for public health.

Material and methods

Study population

This study is a nationwide investigation conducted in 2019 assessing the physical fitness of students aged 6 to 22 from a representative sample across 32 provinces, municipalities, and autonomous regions in China. A total of 827 schools participated in the study, selected from the central, eastern, western, southern, and northern areas of each city to ensure high representativeness. Nationwide, 241,536 students were enrolled, although data from Hubei Province could not be collected due to the COVID-19 pandemic. Students from fourth grade in elementary school through the third grade of high school were asked to participate in a survey regarding their lifestyle and general health.

A total of 161,426 students completed the questionnaire in this study. Of these, 161,112 were aged 9 to 18 years. Boys and girls with a medical problem or who were missing from one of the tests were excluded. The total sample size for this investigation comprised 161,093 participants aged 9–18 years, ranging from the third grade of elementary school to the third grade of high school. The study included a physical examination, an examination of common diseases, and a lifestyle questionnaire. The school teachers explained the survey program to the participants and their parents. A verbal agreement was obtained from the parents of the students who took part in the research. The study was approved by the Ethics Committee of Shaanxi Normal University (202016001).

Demographic characteristics

The survey gathered comprehensive data on demographic parameters, obesity-related behavioral factors, and the physical activity status of adolescents using the Evaluation Indicator System of Physical Activity and Physical Fitness for Youth (EISPAFY) [17]. The survey has been validated in the Physical Fitness and Health Study of Chinese Youth (PAFCTYS). Demographic characteristics included age, sex, and level of parental support for student participation in exercise (very supportive, supportive, neutral, unsupportive, and very unsupportive). The behavioral components associated with obesity were assessed based on the self-reported frequency of occurrence for each participant over the past seven days, including sedentary time (< 8 h, 8–10 h, > 10 h), frequency of breakfast consumption (never, sometimes, every day), and sugar-sweetened beverage intake (never, < 1 time/day, ≥ 1 time/day).

Air pollutants

The outdoor air pollutants (PM_2.5, PM₁₀, SO₂, NO₂, CO and O₃) in this study are sourced from the China High Air Pollutants (CHAP) dataset, a comprehensive, high-resolution, and high-quality dataset of atmospheric pollutants across China. This dataset is generated using big data, including ground measurements, satellite remote sensing products, atmospheric reanalysis, and model simulations, combined through artificial intelligence, taking into account the spatiotemporal heterogeneity of air pollution [18–21].

Similar to previous studies [14, 22], this study first converted the addresses of the 827 schools into latitude and longitude data, then calculated concentrations of PM_2.5, PM₁₀, SO₂, NO₂, CO, and O₃ for each specific location. The annual mean values of air pollutants from 2018 were used as an indicator to assess long-term air pollution exposure, as this year preceded the baseline survey and can be regarded as representative of participants’ prolonged exposure to pollutants, considering the actual conditions of Chinese children and adolescents. The baseline survey, including physical examinations and questionnaires, was conducted from September to December 2019. Ambient air pollution exposure was assessed using 2018 annual mean concentrations, representing long-term exposure prior to the outcome assessment. Additionally, sensitivity tests were conducted using the average data from the past three, four, and five years, matching the school latitude and longitude with the corresponding neighborhood committee or administrative village.

In order to address the mixing influence produced by meteorological conditions, we obtained monthly average meteorological data (including average temperature and relative humidity) from the China Meteorological Data Service Center (http://data.cma.cn/en) for 124 cities located in 31 provinces throughout China. Before publishing on the website, meteorological factors and air pollution data were quality-controlled following nationally unified standards for accuracy and reliability.

Measurements of body weight

Participants underwent a health test in the morning after an overnight fast. The height and weight measurements had a precision of 0.1 cm and 0.1 kg, respectively, when the person was dressed in light clothing and was not wearing footwear. The body mass index (BMI) is calculated by dividing an individual’s weight (kg) by the square of their height (m²). Waist circumference is measured 1 cm above the umbilicus with a precision of 0.1 cm. The waist-to-height ratio (WHtR) is expressed as the ratio of waist circumference (cm) to height (cm). Competent staff performed all measurements following specified procedures, and the equipment was calibrated both before and weekly after data compilation [23]. We used the Chinese Obesity Working Group’s age- and sex-specific BMI criteria (BMI ≥ 95th percentile threshold) to define overweight [24]. WHtR ≥ 0.48 was the criterion for central obesity [25, 26].

Statistical analysis

Following prior research [12, 27], to investigate the connection between extended exposure to ambient air pollution and obesity, we implemented a binary regression model. The independent variables were pollutant concentrations, while the dependent variable (a binary variable) in the regression model was overweight and central obesity. The odds ratio (OR) and its corresponding 95% confidence interval (CI) were computed for every 10 μg/m³ increase in pollutant concentration. Studies have shown that an imbalance between calorie intake and physical activity (e.g., excessive consumption of sugary beverages, too much sedentary time, etc.) are the leading causes of obesity in children and adolescents, and that, in addition to genetic and biological factors, social environmental factors, including families (supportive parents), schools, communities, and national policies, also play a critical role [28–31]. Based on findings from earlier research, the regression model included factors related to obesity and demographics [12, 14–16, 27, 32].

We created three models by progressively incorporating additional factors, including: (1) Model 1 () used age (a numerical variable) and sex (a categorical variable) to modify the generalized linear model. (2) Model 2 () adjusted for parents support for students’ participation in sports (categorical variable), the degree of parents enjoy participating in sports (categorical variable), average daily sedentary time (numeric variable), frequency of eating breakfast (categorical variable), and sugar-sweetened beverage consumption (categorical variable) based on Model 1. (3) Model 3 () was calibrated using the average temperature and average relative humidity figures obtained from Model 2. denotes the probability of overweight or central obesity; is the intercept; are regression coefficients corresponding to each covariate. The terms represent concentrations of air pollutants, including AQI, PM_2.5, PM₁₀, SO₂, NO₂, O₃, and the reciprocal of CO.

Given the high correlations among several air pollutants, a multi-pollutant model was utilized for the final data analysis to better control for co-pollutant confounding and estimate independent effects. Variance Inflation Factor (VIF) values were calculated to assess multicollinearity, and all included pollutants had VIFs below 5, indicating acceptable levels of collinearity. Additionally, single-pollutant models were performed for comparison, and their results are presented in Supplementary Table S4. Moreover, we categorized the data based on sex, educational stages, geographical regions (North and South), urban–rural classification, and economic status (high income, middle income, and low income). The stages of education were grouped into compulsory education (grades 4 to 9, including primary and junior secondary school) and high school education (grades 10 to 12). For geographical stratification, we utilized the established North–South division demarcated by the Qinling-Huaihe River line. This approach is widely adopted in Chinese environmental epidemiology because it effectively captures significant climatic variations, distinct living habits, and differing pollution characteristics, allowing for the analysis of macroscopic regional differences while maintaining the statistical stability of sample sizes. This provides a broader regional perspective than a provincial-level analysis [14, 33]. The economic condition was assessed according to the gross domestic product (GDP) per capita at the regional level. The GDP statistics were acquired from the National Bureau of Statistics of China (http://www.stats.gov.cn).

Additionally, we conducted various sensitivity studies, which involved altering the time windows used to calculate air pollutant concentrations. In the primary analyses, instead of using the average concentrations of 1 year before the baseline survey, we used the average concentrations of 5 years, 4 years, and 3 years before the baseline survey. Additionally, we adjusted for educational stages instead of age in the primary model grades, which may offer a more precise representation of students’ development trajectories and physical activity patterns. Due to the relatively low levels and right-skewed distribution of CO in our dataset, we applied a reciprocal transformation (CO⁻¹) to improve model fit and stabilize variance. This transformation helps address skewness and facilitates the detection of potential inverse associations between CO exposure and overweight and central obesity. Statistical analyses were conducted using R 3.4.3 (R Core Team, Vienna, Austria). A statistically significant value was defined as a two-tailed P value of less than 0.05.

Results

Descriptive analysis

The general characteristics of the study participants are presented in Table 1. The current study included a population sample of 161 093 children and adolescents aged 9–18 years, with a nearly equal distribution of sexes (52.1% girls and 47.9% boys). The average BMI was 19.04 kg/m²(with a standard deviation of 2.97). The frequency of overweight was 10.8% (17 429/161 093), and the occurrence of central obesity was 17.6% (28 344/161 093). For comparison, boys exhibited larger waist circumferences, weights, and heights. Boys had a higher prevalence of overweight (11.7% vs. 10%) compared to girls.

Table 1.

Demographic characteristics of study participants (mean (SD)/n (%))

Variables	Boys (n = 77 236)	Girls (n = 83 857)
Age, year	13.84(2.8)	13.71(2.7)
Educational Stages, n (%)
Elementary school	22 723(29.5)	24 975(29.6)
Middle school	23 283(30.1)	25 991(30.9)
High school	31 230(40.4)	32 891(39.5)
Urban–Rural Classification, n (%)
Urban	38,709(50.1)	42,046(50.1)
Rural	38,527(49.9)	41,811(49.9)
Region
North	34 059(44.1)	37 728(45)
South	43 177(55.9)	46 129(55)
Economic status, n (%)
Low income	25 015(32.4)	24 975(29.8)
Middle income	29 125(37.7)	31 791(37.9)
High income	23 096(29.9)	25 981(32.3)
Weight, kg	49.77(14.07)	45.98(10.61)
Height, cm	160.18(14.54)	154.17(9.84)
Waist circumstance, cm	67.294(9.17)	65.31(7.99)
Body mass index, kg/m2	18.97(2.96)	19.09(2.97)
Waist-to-height ratio	0.42(0.05)	0.42(0.04)
Overweight, n (%)	9 071(11.7)	8 358(10)
Central obesity, n (%)	13 031(16.8)	15 313(18.2)
Average daily sedentary time(h/day), n (%)
< 8 h	28 766(37.2)	29 402(35)
8-10 h	21 179(27.4)	24 823(29.6)
> 10 h	27 291(36.4)	29 632(35.4)
Frequency of eating breakfast, n (%)
Never	4 303(5.6)	4 034(4.8)
Sometimes	18 329(23.7)	20 397(24.3)
Everyday	54 604(70.7)	59 426(70.9)
Sugar-sweetened beverage consumption n (%)
Never	15 766(20.4)	19 479(23.2)
< 1 time/day	41 923(54.3)	47 681(56.9)
≥ 1 time/day	19 547(25.3)	16 697(19.9)
Parents support students’ participation in sports n (%)
Supportive and very supportive	59 971(77.7)	66 287(79.1)
Neutral	14 093(18.2)	15 029(17.9)
Not very supportive and very unsupportive	3 097(4.1)	2 472(3.0)
Parents enjoy participating in sports n (%)
Parents love it	34 546(44.8)	39 791(47.5)
Father loves it	15 547(20.1)	16 393(19.6)
Mother loves it	7 099(9.2)	8 710(10.4)
Parents do not like it	19 965(25.9)	18 891(22.5)

For continuous variables, the data are presented as the mean (standard deviation); for binary variables, they are presented as the number (percentage). Data were collected from September to December 2019

The average distribution of air pollutant concentrations is shown in Table 2. In general, the particle concentrations were far greater than those recommended by the World Health Organization (5 μg/m³ for annual mean PM_2.5, 15 μg/m³ for annual mean PM₁₀, and 10 μg/m³ for annual mean NO₂). The air pollutants, particularly PM_2.5 and PM₁₀, showed a positive statistical correlation (correlation coefficient = 0.938) (Table S1). Figure S1 illustrates the distribution of the AQI across various provinces in China. Regions such as Xinjiang, Beijing, and Hebei have higher AQI values, indicating more severe air pollution. In contrast, regions like Tibet and Hainan exhibit lower AQI values, suggesting better air quality.

Table 2.

Distribution of the 2018 annual average exposure to air pollution and average monthly meteorological parameters in 2018

Air pollutants/meteorological parameters	Mean	Median	Minimum	Maximum	SD	IQR
PM2.5 (μg/m3)	42.88	39.42	9.8	97.52	18.89	21.21
PM10 (μg/m3)	69.63	63.62	15.42	202.43	33.55	32.13
SO2 (μg/m3)	10.46	9.27	2.08	35.79	5.39	6.42
NO2 (μg/m3)	33.36	33.88	10	57.57	14.41	21.47
CO (mg/m3)	0.91	0.86	0.52	1.69	0.26	0.35
O3(μg/m3)	49.28	46.43	24.24	91.57	16.21	16.06
AQI	65.55	60.40	1.97	151.61	28.39	24.64
Temperature (°C)	11.77	12.88	−1.28	23.90	6.67	9.24
Relative humidity (%)	67.23	69.5	19.13	87.25	11.01	14.25

Air pollutant concentrations represent annual averages for 2018, while temperature and relative humidity are averaged monthly values from 2018

Abbreviations: AQI Air Quality Index, PM_2.5 Particulate matter with aerodynamic diameter < 2.5 μm, PM₁₀ Particulate matter with aerodynamic diameter < 10.0 μm, SO₂ Sulfur Dioxide, NO₂ Nitrogen dioxide, O₃ Ozone, CO Carbon Monoxide, SD Standard deviation, IQR Interquartile range

Association between ambient air pollutant exposure and overweight and central obesity

Air pollution and the prevalences of central obesity and overweight are depicted in Figs. 1 and 2, with detailed statistical associations shown in Table S3.

Fig. 1.

Odds ratios for overweight associated with a 10 µg/m³ increase in ambient air pollutant concentration across three models. Abbreviations: AQI, Air Quality Index; PM_2.5, particulate matter with aerodynamic diameter < 2.5 μm; PM₁₀, particulate matter with aerodynamic diameter < 10.0 μm; SO₂, Sulfur Dioxide; NO₂, nitrogen dioxide; O₃, Ozone; CO, Carbon Monoxide. (Note: CO values are represented as the reciprocal of the calculated values). Model 1 was based on the generalized linear model, adjusted for sex (a categorical variable) and age (a numeric variable). Model 2 was adjusted for Parents’ support of students’ participation in sports (categorical variable), the degree to which parents enjoy participating in sports (categorical variable), average daily sedentary time (numeric variable), frequency of eating breakfast (categorical variable), and sugar-sweetened beverage consumption (categorical variable) based on Model 1. Model 3 was adjusted using the mean temperature and mean relative humidity values from Model 2

Fig. 2.

Odds ratios for central obesity associated with a 10 µg/m³ increase in ambient air pollutant concentration across three models. Abbreviations: AQI, Air Quality Index; PM_2.5, particulate matter with aerodynamic diameter < 2.5 μm; PM₁₀, particulate matter with aerodynamic diameter < 10.0 μm; SO₂, Sulfur Dioxide; NO₂, nitrogen dioxide; O₃, Ozone; CO, Carbon Monoxide. (Note: CO values are represented as the reciprocal of the calculated values). Model 1 was based on the generalized linear model, adjusted for sex (a categorical variable) and age (a numeric variable). Model 2 was adjusted for Parents’ support of students’ participation in sports (categorical variable), the degree to which parents enjoy participating in sports (categorical variable), average daily sedentary time (numeric variable), frequency of eating breakfast (categorical variable), and sugar-sweetened beverage consumption (categorical variable) based on Model 1. Model 3 was adjusted using the mean temperature and mean relative humidity values from Model 2

For overweight, PM_2.5, SO₂, NO₂, and CO (reciprocal transformed) were positively associated with increased odds across all three models. Specifically, in the fully adjusted Model 3, each 10 μg/m³ increment in PM_2.5 was associated with 21.2% higher odds of being overweight (OR = 1.212, 95% CI: 1.169–1.257, p < 0.001). Similarly, both SO₂ and NO₂ were associated with increased odds of overweight, with ORs of 1.103 (95% CI: 1.062–1.145) and 1.069 (95% CI: 1.046–1.092), respectively (both p < 0.001), while CO⁻¹ was also positively associated (OR = 1.052, 95% CI: 1.043–1.061, p < 0.001). In contrast, PM₁₀ was inversely associated with overweight (OR = 0.954, 95% CI: 0.938–0.969, p < 0.001), and AQI showed no significant association in the final model (OR = 0.959, 95% CI: 0.935–0.983, p < 0.001). O₃ was negatively associated with overweight (OR = 0.949, 95% CI: 0.935–0.962, p < 0.001).

For central obesity, positive associations were observed with AQI, SO₂, and NO₂ in Model 3, with odds ratios (ORs) of 1.118 (95% CI: 1.095–1.141), 0.980 (not statistically significant), and 1.065 (95% CI: 1.051–1.079), respectively. PM₁₀ showed a consistent inverse association (OR = 0.901, 95% CI: 0.889–0.913, p < 0.001), while PM_2.5 and CO⁻¹ were not significantly associated with central obesity after full adjustment. O₃ also showed a negative association (OR = 0.973, 95% CI: 0.962–0.985, p < 0.001).

Figure S2 presents the distribution of overweight prevalence across provinces in China. Northern provinces such as Beijing, Tianjin, and Hebei show notably high prevalence rates of overweight individuals. In contrast, southern and western provinces, such as Tibet and Xinjiang, exhibit lower prevalence rates. Figure S3 depicts the distribution of central obesity prevalence across Chinese provinces. Shaanxi and Qinghai show notably high prevalence rates of central obesity. Conversely, regions like Guangxi and Jiangsu exhibit lower prevalence rates.

Stratified analysis

Tables 3 and 4 present the results of the stratification analyses. Within each subgroup categorized by sex, educational stage, economic status, and region, noteworthy positive correlations were identified between AQI, PM_2.5, PM₁₀, SO₂, and NO₂ with overweight and central obesity, while negative correlations were found between O₃ and CO⁻¹. In general, the positive relationship between exposure to PM_2.5, PM₁₀, SO₂, and NO₂ and the risk of overweight was more pronounced in boys, low-income regions, and the southern part of the country. Among different school-level subgroups, the risk of overweight associated with exposure to PM_2.5, PM₁₀, SO₂, and NO₂ was most substantial in participants in high school. Additionally, the strongest correlation between exposure to PM_2.5, SO₂, and NO₂ and the risk of central obesity was found in middle-income participants.

Table 3.

Stratified analysis of odds ratios (OR) for overweight associated with a 10 µg/m³ increase in ambient air pollutant concentration across population subgroups

Stratification	PM2.5 OR (95% CI)	PM10 OR (95% CI)	SO2 OR (95% CI)	NO2 OR (95% CI)	CO−1 OR (95% CI)	O3 OR (95% CI)	AQI OR (95%CI)
Urban	1.043 (1.031–1.055)	1.012 (1.005–1.018)	1.198 (1.144–1.253)	1.106 (1.086–1.127)	0.994 (0.986–1.001)	0.945 (0.929–0.961)	1.029 (1.020–1.037)
Rural	1.063 (1.050–1.077)	1.025 (1.018–1.033)	1.055 (1.006–1.106)	1.069 (1.050–1.087)	0.992 (0.984–0.999)	0.917 (0.902–0.932)	1.033 (1.024–1.041)
South	1.076 (1.055–1.098)	1.028 (1.015–1.042)	1.090 (1.019–1.166)	1.070 (1.047–1.093)	0.996 (0.985–1.007)	0.958 (0.943–0.973)	1.042 (1.028–1.055))
North	1.012 (1.000–1.023)	0.998 (0.992–1.004)	1.093 (1.046–1.142)	1.042 (1.021–1.063)	1.005 (0.999–1.012)	0.931 (0.910–0.954)	1.016 (1.009–1.024)
Low income	1.100 (1.081–1.120)	1.059 (1.048–1.071)	1.022 (0.975–1.072)	1.092 (1.067–1.117)	0.969 (0.959–0.980)	0.891 (0.863–0.920)	1.052 (1.039–1.066)
Middle income	0.996 (0.982–1.011)	0.986 (0.978–0.993)	1.133 (1.063–1.207)	1.077 (1.049–1.105)	1.015 (1.007–1.029)	0.959 (0.936–0.983)	0.993 (0.982–1.023)
High income	1.029 (1.007–1.052)	1.001 (0.987–1.015)	0.825 (0.740–0.918)	1.033 (1.008–1.058)	1.001 (0.991–1.010)	0.964 (0.944- 0.984)	1.019 (1.003–1.036)
Compulsory education stage	1.029 (1.016–1.043)	1.010 (1.003–1.017))	1.073 (1.022–1.126)	1.051 (1.032–1.071)	0.985 (0.977–0.993)	0.939 (0.923–0.956)	1.016 (1.007–1.024)
High school	1.064 (1.052–1.076)	1.022 (1.016–1.029)	1.165 (1.113–1.218)	1.103 (1.085–1.122)	0.995 (0.988–1.002)	0.926 (0.911–0.941)	1.040 (1.032–1.048)
Girls	1.041 (1.028–1.053)	1.015 (1.008–1.022)	1.047 (0.999–1.097)	1.053 (1.035–1.072)	0.994 (0.986–1.001)	0.926 (0.910–0.942)	1.022 (1.014–1.031)
Boys	1.064 (1.051–1.076)	1.022 (1.015–1.028)	1.207 (1.154–1.263)	1.117 (1.098–1.137)	0.991 (0.984–0.999)	0.933 (0.918–0.948)	1.039 (1.031–1.047)

(Note: CO values are represented as the reciprocal of the calculated values.) Adjusted for age, sex, economic income level, parents’ support for students’ participation in sports, parents’ enjoyment of participating in sports, average daily sedentary time, frequency of eating breakfast, sugar-sweetened beverage consumption, mean temperature, and mean relative humidity

Abbreviations: AQI Air Quality Index, PM_2.5 Particulate matter with aerodynamic diameter < 2.5 μm, PM₁₀ Particulate matter with aerodynamic diameter < 10.0 μm, SO₂ Sulfur Dioxide, NO₂ Nitrogen dioxide., O₃ Ozone, CO Carbon Monoxide

Table 4.

Stratified analysis of odds ratios (OR) for central obesity associated with a 10 µg/m³ increase in ambient air pollutant concentration across population subgroups

Stratification	PM2.5 OR (95% CI)	PM10 OR (95% CI)	SO2 OR (95% CI)	NO2 OR (95% CI)	CO−1 OR (95% CI)	O3 OR (95% CI)	AQI OR (95%CI)
Urban	0.997 (0.988–1.007)	0.990 (0.984–0.995)	1.022 (0.985–1.061)	1.037 (1.022–1.053)	0.988 (0.982–0.994)	0.981 (0.966–0.995)	1.001 (0.995–1.008))
Rural	1.020 (1.010–1.030)	0.999(0.994–1.005)	1.003 (0.967–1.041)	1.052 (1.038–1.067)	0.998 (0.992–1.004)	0.922 (0.909–0.934)	1.016 (1.009–1.023)
South	1.073 (1.055–1.091)	1.016 (1.005–1.028)	1.068 (1.010–1.129)	1.048 (1.029–1.067)	0.998 (0.989–1.008)	0.910 (0.897–0.921)	1.054 (1.043–1.066)
North	0.968 (0.959–0.977)	0.976 (0.971–0.981)	0.941 (0.910–0.975)	1.013 (0.998–1.029)	1.002 (0.997–1.007)	1.052 (1.034–1.071)	0.973 (0.967–0.979)
Low income	1.028 (1.014–1.043)	1.021 (1.012–1.031)	1.104 (1.065–1.144))	1.082 (1.064–1.101)	0.995 (0.986–1.004)	0.94 (0.91–0.97)	1.053 (1.043–1.064)
Middle income	0.984 (0.972–0.995)	0.977 (0.971–0.983)	0.981 (0.931–1.034)	1.077 (1.054–1.101)	1.001 (0.992–1.010)	0.878 (0.859–0.895)	0.978 (0.969–0.986)
High income	0.974 (0.955–0.993)	0.959 (0.947–0.971)	0.850 (0.788–0.917)	0.969 (0.949–0.989)	1.028 (1.019–1.036)	0.969 (0.952–0.986)	0.975 (0.961–0.989)
Compulsory education stage	0.998 (0.989–1.008)	0.991 (0.986–0.996)	1.030 (0.995–1.067)	1.048 (1.034–1.063)	0.988 (0.982–0.995)	0.949 (0.936–0.962)	1.004 (0.998–1.010)
High school	1.016 (1.005–1.026)	0.997 (0.992–1.003)	1.004 (0.966–1.044)	1.051 (1.036–1.066)	0.998 (0.992–1.004)	0.951 (0.938–0.965)	1.012 (1.006–1.019)
Girls	1.017 (1.007–1.026)	1.002 (0.997–1.007)	1.113 (1.076–1.151)	1.043 (1.029–1.058)	0.985 (0.979–0.990)	0.929 (0.917–0.942)	1.012 (1.006–1.019)
Boys	0.998 (0.988–1.008)	0.986 (0.981–0.992)	1.044 (1.005–1.085)	1.045 (1.030–1.061)	1.005 (0.999–1.012)	0.972 (0.958–0.985)	1.004 (0.998–1.011)

(Note: CO values are represented as the reciprocal of the calculated values.) Adjusted for age, sex, economic income level, parents’ support for student’s participation in sports, parents’ enjoyment of participating in sports, average daily sedentary time, frequency of eating breakfast, sugar-sweetened beverage consumption, mean temperature, and mean relative humidity

Abbreviations: AQI Air Quality Index, PM_2.5 Particulate matter with aerodynamic diameter < 2.5 μm, PM₁₀ Particulate matter with aerodynamic diameter < 10.0 μm, SO₂ Sulfur Dioxide, NO₂ Nitrogen dioxide., O₃ Ozone, CO Carbon Monoxide

Sensitivity analysis

Our sensitivity studies produced comparable results. Furthermore, the use of average concentrations of air pollutants across three, four, and five years also produced consistent findings (Table S2).

Discussion

In this large-scale cross-sectional study of Chinese children and adolescents, we found that exposure to specific ambient air pollutants was significantly associated with an increased risk of overweight and central obesity. Specifically, higher levels of PM_2.5, NO₂, SO₂, and CO (inverted) were consistently associated with a greater likelihood of being overweight across all models. For central obesity, NO₂ and AQI remained positively associated after full adjustment. In contrast, PM₁₀ and O₃ were inversely associated with both overweight and central obesity. These findings were robust across multiple models and remained consistent after adjusting for demographic, behavioural, and meteorological confounders. The patterns suggest that specific pollutants have distinct effects on obesity phenotypes, highlighting the complexity of air pollution’s contribution to childhood overweight and central obesity. Factors such as age, sex, sedentary behaviours, and economic status influenced these correlations. Generally, boys exhibited stronger associations compared to girls, except for central obesity, supporting the independent health effects of particulate matter. When adjusting for demographic and behavioural confounders, the observed associations between air pollution and central obesity were generally attenuated, suggesting that factors such as age, physical activity, and parental support may partly confound the relationship between air pollution and central obesity. However, after further adjusting for meteorological variables, including temperature and humidity, some associations became stronger, indicating that meteorological conditions may modify or even amplify the health effects of air pollutants rather than simply acting as confounders. These findings underscore the importance of considering both behavioural and environmental factors when assessing the independent contribution of air pollution.

Environmental pollution affects the endocrine system, contributing to obesity, with each pollutant having distinct impacts. Children and adolescents are particularly vulnerable due to their critical growth stages and underdeveloped organs and physiological systems. Recent epidemiological studies, primarily in North America and Europe, examined the relationship between air pollution and childhood obesity. The Southern California Children’s Health Study, which followed 3,318 children aged 10 to 18, found a significant association between high residential traffic density and nitrogen oxide (NO_x) exposure with increased BMI, suggesting that prolonged exposure to traffic pollution may lead to greater weight gain among children [8, 34, 35]. In another Los Angeles-based study, a cohort of 314 high-risk Latino children showed that long-term exposure to NO₂ and PM_2.5 was correlated with increased BMI and impaired glucose metabolism, linking air pollution to early metabolic disruptions and an increased risk of obesity [7]​. In Europe, the Dutch PIAMA study followed a large sample of children and found a positive association between NO₂ exposure and childhood overweight [36]​. Similarly, a cross-sectional study in Spain involving 2,660 children aged 7 to 10 found that exposure to high levels of particulate matter and ultrafine particles around schools increased the likelihood of being overweight or obese, underscoring the role of environmental exposure in both home and school settings in determining obesity risk.

Nevertheless, other investigations have documented contrasting findings. An Italian study of 719 adolescents found no significant association between prolonged exposure to PM_2.5, PM₁₀, and NO₂ and BMI, waist circumference, or overweight prevalence, possibly due to the smaller sample size or specific regional pollution characteristics ​ [13]. Another investigation revealed no statistically significant relationship between exposure to traffic density, black carbon, or PM_2.5 and childhood obesity among 1,418 children in the United States [37].

While many studies have yielded similar findings, some have not established a significant link between air pollution and childhood obesity. These discrepancies may be due to smaller sample sizes, regional variations in pollution levels, and environmental factors. With a large sample size, extensive geographic coverage, and diverse population, our study addresses these limitations, providing a more comprehensive exploration of the association between air pollution and overweight and central obesity in children and adolescents.

Various biochemical mechanisms are frequently cited as primary causal factors linking air pollution to obesity, although much remains to be learned about these mechanisms [8, 38]. From a metabolic perspective, air contaminants inhaled through the respiratory tract can elevate oxidative stress in tissues and systems [6]. PM_2.5 can impact the expression of genes in the mitochondria of brown adipose tissue, leading to an increase in the formation of reactive oxygen species within this tissue. This can lead to metabolic dysfunction and increase the risk of lipid and glucose metabolism disorders [39, 40]. Furthermore, the presence of air pollution can initiate an inflammatory response that can result in harm to blood vessels and hinder the body’s ability to respond to insulin, impacting body weight [39, 41]. Finally, behavioral mechanisms can also explain this relationship. For example, air pollution reduces the willingness to participate in outdoor activities [42] and increases the consumption of trans-fats and fast food [43], which may contribute to obesity.

Crucial to mitigating the adverse health effects of air pollution is the identification of susceptible subpopulations [14, 44, 45].. Sex-specific characteristics, such as hormone levels and body size, can influence the transfer of chemicals generated by the surrounding environment, including air pollution. This can result in distinct physiological reactions [46]. The varying health impacts of air pollution on different sex populations may also be attributed to social and behavioural factors. Boys in our study exhibited a greater amount of time engaged in outdoor activities compared to girls, increasing their exposure to air pollution. Limited research has investigated the impact of gender on the correlation between air pollution and childhood obesity, yielding inconclusive results [7–9, 35, 37]. More research is needed to explore sex differences in the obesogenic effects of air pollution.

Furthermore, stratified analyses revealed that individuals residing in low-income areas faced an elevated risk of developing obesity due to air pollution, consistent with the findings of Liu et al. (2020). Individuals with lower socioeconomic status may experience higher stress levels, increasing stress hormones that contribute to fat accumulation and obesity [47]. Additionally, age differences may also play a role in the varying impacts of air pollution on the risk of overweight and central obesity. Our stratified analysis revealed that the correlation between exposure to PM2.5, PM10, SO2, and NO2 and the risk of being overweight was strongest among high school students. It may be because adolescents in high school are at a critical stage of physical and hormonal development, making them more susceptible to environmental stressors, such as air pollution. During this period, hormonal changes—including fluctuations in cortisol and insulin sensitivity—may interact with external pollutants to exacerbate weight gain and fat accumulation [48].

Furthermore, high school students often face academic pressures, which may lead to lower physical activity levels compared to younger age groups. This combination of physiological and behavioral factors could increase the risk of overweight and central obesity in response to pollution exposure [49]. To reduce the societal burden of childhood and adolescent overweight and central obesity, future strategies should emphasize both environmental and behavioural interventions. Improving air quality, promoting physical activity, and tailoring public health campaigns to vulnerable subpopulations—such as those in low-income or high-pollution areas—may be practical approaches. These efforts require coordinated actions across public health, education, and environmental sectors.

Regional differences may also contribute to the observed variations in pollution-related obesity risks. In southern China, the stronger positive association between air pollution and obesity might be influenced by climate and lifestyle factors. The warmer and more humid climate in southern regions could exacerbate the effects of air pollution by promoting pollutant retention in the atmosphere, making outdoor activities less appealing and thereby decreasing physical activity levels among children and adolescents [50]. Additionally, individuals in these regions may spend more time indoors, where ventilation is often limited, resulting in prolonged exposure to both indoor and outdoor pollutants. These environmental and behavioral factors could amplify the obesogenic effects of air pollution in the southern part of the country [51]. Temperature and relative humidity are important meteorological factors that can influence both the concentration and toxicity of ambient air pollutants. Higher temperatures can enhance photochemical reactions, increasing the formation of secondary pollutants such as ozone [52]. Relative humidity affects particulate matter through hygroscopic growth and changes in chemical composition, which can alter particle size distribution, deposition patterns in the respiratory tract, and thus toxicity [53].

However, the precise mechanisms through which warmer and more humid conditions modulate pollutant toxicity and related health impacts remain complex and incompletely understood. In this study, temperature and humidity were adjusted for as confounders, but further mechanistic research is needed to clarify their role in modifying the health effects of air pollution.

This study offers several benefits. Firstly, it encompassed a significant number of children and adolescents from diverse areas with various levels of air pollution, creating a large nationally representative sample and providing a wealth of information on various potential confounders. Such development enabled us to provide a more accurate description of the correlation between air pollution and childhood obesity and overweight. Secondly, this research is among the few in China that specifically examines the impacts of air pollution on overweight and central obesity in children and adolescents, extending the focus beyond adult populations. Thirdly, unlike the majority of earlier research that measured overweight solely using BMI, we added WHtR and waist circumference as indicators of central obesity. Regardless of total body obesity, research has established a correlation between central obesity and an increased susceptibility to cardiovascular disease in both adults and children. Therefore, this provides an additional means to evaluate the impact of air pollution on childhood obesity comprehensively.

Our investigation is subject to certain limitations. First, the cross-sectional design limits our capacity to demonstrate causality. Secondly, air pollutant concentrations were estimated based on school addresses. Residential addresses were unattainable owing to insufficient data. Nevertheless, children devote a significant amount of time to school, notably during the mornings when there is a peak in ambient air pollution [54]. Due to China’s school enrollment policies, school-age participants typically live close to their schools. Therefore, school-based estimates provide a partial representation of participants’ prolonged exposure to ambient air pollution [45, 55]. Third, the model’s covariates were not comprehensive enough due to the limited number of behavioral questions related to potential causes of obesity included in the questionnaire. Important behavioural factors, such as detailed physical activity patterns, dietary habits beyond sugar-sweetened beverage intake, screen time, sleep duration, and psychosocial stress, were not assessed in our study. Incorporating these additional behavioural variables in future research could improve adjustment for confounding and provide deeper insights into the pathways linking air pollution exposure to overweight and central obesity. Future studies should combine residential and school addresses for comprehensive exposure estimates. Fourth, one limitation is that not all Chinese provinces were included due to missing or incomplete data. This may limit the representation of some regions. However, the included provinces span diverse geographic and environmental contexts, supporting the overall generalizability of the findings. Finally, limited direct comparison of results between different air pollutants arises from the employment of several models with varying prediction abilities to forecast air pollutants. Moreover, although this study adopted the well-established North–South China division based on the Qinling Mountains–Huaihe River Line to reflect significant geographic and climatic differences, this binary classification cannot fully capture finer spatial heterogeneity.

Future research should leverage province-level or even county-level data, combined with high-resolution satellite and monitoring network exposures, employing multilevel mixed-effects models to explore more granular regional risk variations. Such refined analyses would enhance understanding of localized exposure-health relationships and inform targeted interventions.

Conclusions

Long-term exposure to air pollution (PM_2.5, PM₁₀, SO₂, NO₂, and CO) was positively associated with the prevalence of overweightness and central obesity in Chinese children and adolescents. Higher levels of these pollutants were significantly correlated with increased rates of overweight and central obesity. Our research suggests that air pollution may contribute to the rising prevalence of overweight and central obesity in children and adolescents.

These findings underscore the importance of implementing and strengthening environmental policies aimed at reducing air pollution levels, particularly in regions with high exposure to pollution. Improved air quality not only supports environmental sustainability but could also play a vital role in mitigating the risks of childhood obesity and related health issues. By addressing air pollution as a contributing factor, public health policies can more effectively target childhood obesity and promote healthier developmental outcomes.

Abbreviations

BMI

Body mass index

WHtR

Waist-to-height ratio

WC

Waist circumference

SBP

Systolic blood pressure

DBP

Diastolic blood pressure

BP

Blood pressure

CI

Confidence interval

OR

Odds ratio

Ref

Reference

AQI

Air Quality Index

CHAP

China High Air Pollutants

CO

Carbon Monoxide

NO₂

Nitrogen Dioxide

O₃

Ozone

PAFCTYS

Physical Fitness and Health Study of Chinese Youth

PM_2.5

Particulate Matter with aerodynamic diameter < 2.5 µm

PM₁₀

Particulate Matter with aerodynamic diameter < 10 µm

SO₂

Sulfur Dioxide

WHtR

Waist-to-Height Ratio

Authors’ contributions

Jiaming Yan：Writing – review & editing, Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Conceptualization. Weixin Chen：Formal analysis. Haiqi Zeng：Visualization, Yanbing Li：Formal analysis, Investigation. Shutong Yang：Formal analysis, Investigation. Longhai Zhang Visualization, Methodology, Investigation, Formal analysis, Conceptualization. Zhenxing Kong：Methodology, Investigation. Yuliang Sun：Investigation, Formal analysis. Wenfei Zhu：Writing – review & editing, Writing – original draft, Visualization, Methodology.

Funding

The study was supported by the General Project of Education, National Social Science Foundation of China (Grant No. BLA230108).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study was approved by the Shaanxi Normal University ethics committee (201916019). Informed consent was obtained from all subjects involved in the study. All experiments were conducted following the Declaration of Helsinki and relevant national and institutional guidelines and regulations.

Competing interests

The authors declare no competing interests.